When can I assume the sampling distribution of x̄ is Normal?

In two cases: if the population itself is Normal (then x is Normal for any n), or if the sample size is large enough for the Central Limit Theorem — commonly n ≥ 30 — even when the population is skewed or unknown.

What is the difference between σ and σ/√n?

is the standard deviation of individual data values in the population. /sqrt(n) is the standard error — the standard deviation of the sample mean x. Use for probabilities about one observation and /sqrt(n) for probabilities about a sample mean.

Why do variances add for the difference of two sample means?

Because the two samples are independent, their variances combine additively regardless of whether you add or subtract the means. So _x_1-x_2 = sqrt(_1^2/n_1 + _2^2/n_2). You add the squared standard errors, then take the square root — never subtract.

Does a larger sample make the population data more Normal?

No. The CLT affects only the sampling distribution of the statistic x, not the raw population data. The population keeps its original shape; it is the distribution of sample means that becomes more Normal as n increases.

AP-STATS-5.7-5.8

U5.7 Sampling Distribution of x̄

Master the sampling distribution of x̄: find its mean and standard deviation, apply the Central Limit Theorem, compute probabilities, and extend to two-sample differences.

What you'll do in this lesson

A voice-first session with the Crimsora tutor on U5.7 Sampling Distribution of x̄, then targeted practice and FRQs — with the tutor adapting to where you get stuck.

Start U5.7 Sampling Distribution of x̄ free

What this lesson covers

Every time you take a sample and compute its mean

\bar{x}

, you get a slightly different value. Collect all those possible sample means and you have a sampling distribution — a distribution of a statistic, not of raw data. This lesson pins down exactly what that distribution looks like: where it centers, how spread out it is, and when it becomes bell-shaped.

Once you know the mean, standard deviation, and shape of the distribution of

\bar{x}

, you can convert questions about sample means into ordinary normal-curve probability problems. We finish by extending everything to the difference of two independent sample means,

\bar{x}_1 - \bar{x}_2

, which sets up the two-sample inference you'll meet in later units.

Center and Spread of the Distribution of x̄

Suppose a population has mean

\mu

and standard deviation

\sigma

. If you draw a simple random sample of size

n

and compute the sample mean

\bar{x}

, then across all possible samples:

\mu_{\bar{x}} = \mu \qquad \sigma_{\bar{x}} = \frac{\sigma}{\sqrt{n}}

The mean of the sampling distribution equals the population mean — this is what makes

\bar{x}

an unbiased estimator of

\mu

. The standard deviation of

\bar{x}

, often called the standard error, is the population standard deviation divided by

\sqrt{n}

.

Notice the

\sqrt{n}

in the denominator: larger samples produce sample means that cluster more tightly around

\mu

. Quadrupling the sample size only halves the standard error, because

\sqrt{4}=2

. This diminishing-returns relationship is a favorite exam point.

The formula

\sigma_{\bar{x}} = \sigma/\sqrt{n}

assumes independent observations. As a rule, this holds when sampling with replacement, or when the sample is less than 10% of the population (the 10% condition). If you sample a large fraction of a finite population without replacement, the true standard error is smaller than the formula suggests.

Shape and the Central Limit Theorem

The shape of the distribution of

\bar{x}

depends on two things: the shape of the population and the sample size.

Situation	Shape of distribution of $\bar{x}$
Population is Normal	Exactly Normal for any $n$
Population not Normal, small $n$	Similar to population (skewed)
Population not Normal, large $n$	Approximately Normal (CLT)

If the population itself is Normal, then

\bar{x}

is Normal no matter how tiny the sample. When the population is skewed or unknown, the Central Limit Theorem guarantees the distribution of

\bar{x}

becomes approximately Normal as

n

grows. The common working threshold on the AP exam is

n \geq 30

.

A frequent misconception: the CLT does not change the shape of the population data itself, and it does not require the sample size to be large relative to the population. It describes only the shape of the sampling distribution of the statistic. Another trap — with a small sample from a heavily skewed population, you cannot assume Normality, so probability calculations based on the Normal model are not justified.

Computing Probabilities with z-scores

Once you know

\bar{x}

is (at least approximately) Normal with mean

\mu_{\bar{x}}

and standard deviation

\sigma_{\bar{x}}

, probability questions become standardization problems. The key formula is:

z = \frac{\bar{x} - \mu_{\bar{x}}}{\sigma_{\bar{x}}} = \frac{\bar{x} - \mu}{\sigma/\sqrt{n}}

The single most common mistake students make is using

\sigma

in the denominator instead of

\sigma/\sqrt{n}

. That confuses the probability that one individual exceeds a value with the probability that a sample mean exceeds it. Because

\bar{x}

has a smaller spread than individual values, sample means rarely land as far from

\mu

as single observations do.

Workflow for these problems: (1) State the shape — is it Normal because the population is Normal, or approximately Normal by the CLT? (2) Compute

\mu_{\bar{x}}

and

\sigma_{\bar{x}}

. (3) Standardize to a

z

-score. (4) Use the standard Normal table or calculator (normalcdf) to find the area. Always check conditions before invoking Normality; on free-response questions, skipping the shape justification costs points.

Difference of Two Sample Means

For two independent samples with means

\bar{x}_1

and

\bar{x}_2

, the sampling distribution of the difference

\bar{x}_1 - \bar{x}_2

has:

\mu_{\bar{x}_1 - \bar{x}_2} = \mu_1 - \mu_2

\sigma_{\bar{x}_1 - \bar{x}_2} = \sqrt{\frac{\sigma_1^2}{n_1} + \frac{\sigma_2^2}{n_2}}

The means subtract, but the variances add — this is because the two samples are independent, and variances of independent random variables always add regardless of whether you sum or subtract them. A frequent error is subtracting the variances or subtracting the standard deviations; never do either.

Do not add standard deviations directly. You must square each standard error, add, then take the square root. The difference

\bar{x}_1 - \bar{x}_2

is Normal if both populations are Normal, or approximately Normal when both sample sizes are large enough for the CLT to apply to each group.

To find a probability about the difference, standardize the same way:

z = \frac{(\bar{x}_1 - \bar{x}_2) - (\mu_1 - \mu_2)}{\sqrt{\frac{\sigma_1^2}{n_1} + \frac{\sigma_2^2}{n_2}}}

Key terms

Sampling distribution of x̄.: The distribution of the sample mean $\bar{x}$ over all possible samples of a fixed size $n$ from a population.
Standard error of the mean.: The standard deviation of the sampling distribution of $\bar{x}$ , equal to $\sigma/\sqrt{n}$ .
Unbiased estimator.: A statistic whose sampling distribution is centered at the parameter it estimates; $\bar{x}$ is unbiased for $\mu$ because $\mu_{\bar{x}}=\mu$ .
Central Limit Theorem (CLT).: The principle that the sampling distribution of $\bar{x}$ becomes approximately Normal as sample size increases, regardless of population shape.
10% condition.: Independence for the standard error formula is reasonable when the sample is less than 10% of the population size.
Normal approximation.: Modeling the distribution of $\bar{x}$ as Normal, justified either by a Normal population or by a large sample via the CLT.
Standardizing (z-score).: Converting a value of $\bar{x}$ to $z=(\bar{x}-\mu)/(\sigma/\sqrt{n})$ so probabilities can be read from the standard Normal distribution.

Worked example

The weights of oranges from an orchard are Normally distributed with mean

\mu = 220

grams and standard deviation

\sigma = 30

grams. A packer randomly selects 36 oranges. What is the probability that the mean weight of the sample exceeds 228 grams?

Step 1: Identify the shape. The population is Normal, so the sampling distribution of

\bar{x}

is exactly Normal for any sample size, including

n=36

.

Step 2: Find the mean and standard deviation of

\bar{x}

. The center is

\mu_{\bar{x}} = \mu = 220

grams. The standard error is

\sigma_{\bar{x}} = \frac{\sigma}{\sqrt{n}} = \frac{30}{\sqrt{36}} = \frac{30}{6} = 5

grams.

Step 3: Standardize the value 228.

z = \frac{228 - 220}{5} = \frac{8}{5} = 1.6

Step 4: Find the probability. We want

P(\bar{x} > 228) = P(z > 1.6)

. From the standard Normal table,

P(z < 1.6) = 0.9452

, so

P(z > 1.6) = 1 - 0.9452 = 0.0548

.

There is about a 0.0548 probability that the mean weight of 36 randomly selected oranges exceeds 228 grams. Notice how using

\sigma_{\bar{x}}=5

rather than

\sigma=30

made the sample mean far more likely to stay near 220.

Practice questions

A population has mean

\mu = 50

and standard deviation

\sigma = 12

. For samples of size

n = 9

, what is the standard deviation of the sampling distribution of

\bar{x}

$12$
$4$
$1.33$
$36$

Answer: $4$

The standard deviation of

\bar{x}

is the standard error

\sigma/\sqrt{n} = 12/\sqrt{9} = 12/3 = 4

. Choosing 12 forgets to divide by

\sqrt{n}

; 1.33 wrongly divides by

n

instead of

\sqrt{n}

Two independent samples are taken. Sample 1 has

n_1 = 40

from a population with

\sigma_1 = 10

; sample 2 has

n_2 = 50

from a population with

\sigma_2 = 8

. Find the standard deviation of the sampling distribution of

\bar{x}_1 - \bar{x}_2

Answer: Approximately 1.96

Variances add for independent samples:

\sigma_{\bar{x}_1-\bar{x}_2} = \sqrt{\frac{\sigma_1^2}{n_1}+\frac{\sigma_2^2}{n_2}} = \sqrt{\frac{100}{40}+\frac{64}{50}} = \sqrt{2.5+1.28} = \sqrt{3.78} \approx 1.94

. (Rounding gives about 1.94–1.96.) Never subtract the variances or add the standard deviations directly.

A skewed population has mean

\mu = 15

and standard deviation

\sigma = 6

. Explain whether you can find

P(\bar{x} > 16)

using a Normal model for a sample of size

n = 4

, and for a sample of size

n = 64

Answer: For n = 4 you cannot; for n = 64 you can.

With

n=4

from a skewed population, the sample size is too small for the CLT to apply, so the distribution of

\bar{x}

is still skewed and the Normal model is not justified. With

n=64 \geq 30

, the CLT makes the distribution of

\bar{x}

approximately Normal, so you may standardize using

\sigma_{\bar{x}} = 6/\sqrt{64} = 0.75

and compute the probability.

FAQ

When can I assume the sampling distribution of x̄ is Normal?: In two cases: if the population itself is Normal (then $\bar{x}$ is Normal for any $n$ ), or if the sample size is large enough for the Central Limit Theorem — commonly $n \geq 30$ — even when the population is skewed or unknown.
What is the difference between σ and σ/√n?: $\sigma$ is the standard deviation of individual data values in the population. $\sigma/\sqrt{n}$ is the standard error — the standard deviation of the sample mean $\bar{x}$ . Use $\sigma$ for probabilities about one observation and $\sigma/\sqrt{n}$ for probabilities about a sample mean.
Why do variances add for the difference of two sample means?: Because the two samples are independent, their variances combine additively regardless of whether you add or subtract the means. So $\sigma_{\bar{x}_1-\bar{x}_2} = \sqrt{\sigma_1^2/n_1 + \sigma_2^2/n_2}$ . You add the squared standard errors, then take the square root — never subtract.
Does a larger sample make the population data more Normal?: No. The CLT affects only the sampling distribution of the statistic $\bar{x}$ , not the raw population data. The population keeps its original shape; it is the distribution of sample means that becomes more Normal as $n$ increases.

Learn this with a teacher, not a page

The Crimsora tutor teaches U5.7 Sampling Distribution of x̄ live — explaining on a whiteboard, asking you questions, and adapting to where you get stuck.

Try a live session free